Skip to content

Instantly share code, notes, and snippets.

@NTT123
Last active July 4, 2018 07:53
Show Gist options
  • Save NTT123/a46e445d31afc870a6ce6ce23d9da887 to your computer and use it in GitHub Desktop.
Save NTT123/a46e445d31afc870a6ce6ce23d9da887 to your computer and use it in GitHub Desktop.
MADE.ipynb
Display the source blob
Display the rendered blob
Raw
{
"nbformat": 4,
"nbformat_minor": 0,
"metadata": {
"colab": {
"name": "MADE.ipynb",
"version": "0.3.2",
"provenance": [],
"collapsed_sections": [],
"include_colab_link": true
},
"kernelspec": {
"name": "python3",
"display_name": "Python 3"
},
"accelerator": "GPU"
},
"cells": [
{
"cell_type": "markdown",
"metadata": {
"id": "view-in-github",
"colab_type": "text"
},
"source": [
"[View in Colaboratory](https://colab.research.google.com/gist/NTT123/a46e445d31afc870a6ce6ce23d9da887/made.ipynb)"
]
},
{
"metadata": {
"id": "Xw1qTgnmTKae",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"## MADE: Masked Autoencoder for Distribution Estimation"
]
},
{
"metadata": {
"id": "_zg9cubEXumS",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"## Method"
]
},
{
"metadata": {
"id": "UeHLtT5PVdf-",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"Let $D$ be the dataset of $n$ instances $D = \\left\\{x^{(1)}, \\dots, x^{(n)}\\right\\}$. \n",
"\n",
"The goal of distribution estimation is to learn a function \n",
"\n",
"$$f: \\mathcal{R}^d \\to [0, 1]$$\n",
"\n",
"which outputs the probability (density) of an instance. It is required that:\n",
"\n",
"$$\n",
"\\int_{x} f(x)~dx = 1\n",
"$$\n",
"\n",
"\n",
"One way to design function $f(x)$, $x = [x_1, \\dots, x_d]$, is to use the chain rule:\n",
"\n",
"$$\n",
"\\begin{align}\n",
"f(x) &= f(x_d \\mid x_{1:d-1}) \\cdot f(x_{1:d-1}) \\\\\n",
"& = f(x_d \\mid x_{1:d-1}) \\cdot f(x_{d-1} \\mid x_{1:d-2}) \\cdots f(x_1)\n",
"\\end{align}\n",
"$$\n",
"\n",
"Here, we use an autoencoding neural network with a $d$-dimensional input which outputs $d$ conditional probabilities.\n",
"However, we have to guarantee that the condional probabilites only depend on its conditional inputs.\n",
" We will use masks to hide connections inside our neural network which violate the constaints. \n",
" \n",
" So our goal here is to design such masks. Considering a single layer neural network:\n",
" \n",
"$$h(x) = g( W \\cdot x + b)$$\n",
"\n",
"where $W$ is the weight matrix, $b$ is the bias vector and $g$ is an element-wise nonlinear function e.g, sigmoid or tanh. We want to design a matrix $M$,\n",
"\n",
"$$h(x) = g( (W \\odot M) \\cdot x + b)$$\n",
"\n",
"which will guarantee the conditional constraint. $\\odot$ is the element-wise multiplication operator.\n",
"\n",
"The solution is assigning each hidden neuron with a random number in $\\{1, \\dots, d-1\\},$ and each input / output neuron a number in range $\\{1, \\dots, d\\}$:\n",
"\n",
" * For a hidden neuron, it will only receive input from neurons in the previous layer whose numbers are **less than or equal to** its number. \n",
" * For an output neuron, it will receive input from neurons whose number are **less than** its number. \n",
" \n",
" \n",
"From these rules, it is guaranteed that the conditional constraint are hold."
]
},
{
"metadata": {
"id": "e9fQjj3mZ2tY",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"<img width=\"412\" alt=\"Screen-Shot-2018-07-03-at-12.28.30-PM\" src=\"data:image/png;base64,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\">"
]
},
{
"metadata": {
"id": "8UgnATb6XzC6",
"colab_type": "text"
},
"cell_type": "markdown",
"source": [
"## Experiments"
]
},
{
"metadata": {
"id": "7fU1co80O6Vw",
"colab_type": "code",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 266
},
"outputId": "4c8ad47c-9a4a-4b8a-8c90-95fd76c639e8"
},
"cell_type": "code",
"source": [
"#@title Load data ...\n",
"# http://pytorch.org/\n",
"from os import path\n",
"from wheel.pep425tags import get_abbr_impl, get_impl_ver, get_abi_tag\n",
"platform = '{}{}-{}'.format(get_abbr_impl(), get_impl_ver(), get_abi_tag())\n",
"\n",
"accelerator = 'cu80' if path.exists('/opt/bin/nvidia-smi') else 'cpu'\n",
"\n",
"!pip install -q http://download.pytorch.org/whl/{accelerator}/torch-0.4.0-{platform}-linux_x86_64.whl torchvision > /dev/null 2>&1\n",
"!wget https://github.com/mgermain/MADE/releases/download/ICML2015/binarized_mnist.npz > /dev/null 2>&1\n",
"import torch\n",
"import numpy as np\n",
"import matplotlib.pyplot as plt\n",
"\n",
"from pylab import rcParams\n",
"rcParams['font.size'] = 14\n",
"plt.style.use('seaborn-white')\n",
"\n",
"#@title\n",
"npzfile = np.load(\"binarized_mnist.npz\")\n",
"train_data = torch.Tensor(npzfile['train_data'])\n",
"valid_data = torch.Tensor(npzfile['valid_data'])\n",
"test_data = torch.Tensor(npzfile['test_data'])\n",
"\n",
"size = 8\n",
"rcParams['figure.figsize'] = size,2*2\n",
"for i in range(1, size+1):\n",
" plt.subplot(2, size//2, i)\n",
" _ = plt.imshow( train_data[i].reshape(28,28) )\n"
],
"execution_count": 0,
"outputs": [
{
"output_type": "display_data",
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAAAd8AAAD5CAYAAABmgj/HAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
"text/plain": [
"<matplotlib.figure.Figure at 0x7f0000fcf940>"
]
},
"metadata": {
"tags": []
}
}
]
},
{
"metadata": {
"id": "9HjkRhT2t0iL",
"colab_type": "code",
"colab": {}
},
"cell_type": "code",
"source": [
"import torch.nn.functional as F\n",
"device = torch.device(0)\n",
"\n",
"class MaskLayer(torch.nn.Module):\n",
" def __init__(self, inputDim, outputDim, nonlinear):\n",
" super(MaskLayer, self).__init__()\n",
" self.inputDim = inputDim\n",
" self.outputDdim = outputDim\n",
"\n",
" self.linear = torch.nn.Linear(inputDim, outputDim)\n",
" self.nonlinear = nonlinear\n",
" \n",
" self.register_buffer('M', \n",
" torch.zeros(outputDim, inputDim).to(device))\n",
"\n",
" self.register_parameter('U', \n",
" torch.nn.Parameter(torch.randn(outputDim, inputDim).to(device)))\n",
" \n",
" def setMask(self, mask):\n",
" self.M.data.copy_(mask)\n",
" \n",
" def forward(self, x):\n",
" out = F.linear(x, self.linear.weight * self.M, self.linear.bias)\n",
" ## out = out + torch.sum(self.U * self.M, dim=1)\n",
" if self.nonlinear is None:\n",
" return out\n",
" return self.nonlinear(out)"
],
"execution_count": 0,
"outputs": []
},
{
"metadata": {
"id": "RqwNtrHsv33q",
"colab_type": "code",
"colab": {}
},
"cell_type": "code",
"source": [
""
],
"execution_count": 0,
"outputs": []
},
{
"metadata": {
"id": "gz_wQ1HSvM7o",
"colab_type": "code",
"colab": {}
},
"cell_type": "code",
"source": [
"class MADE(torch.nn.Module):\n",
" \n",
" def __init__(self, hidSize = 500, numMask=1):\n",
" super(MADE, self).__init__() \n",
"\n",
" self.hidSize = hidSize\n",
" self.numMask = numMask\n",
"\n",
" self.hidden = MaskLayer(28*28, self.hidSize, torch.nn.ELU())\n",
" self.output = MaskLayer(self.hidSize, 28*28, None)\n",
" self.loss = torch.nn.BCEWithLogitsLoss(reduce=False)\n",
" \n",
" self.masks = []\n",
" for i in range(numMask):\n",
" self.masks.append( self.generateMask() )\n",
" \n",
" self.currentMask = -1\n",
" self.nextMask()\n",
"\n",
" def generateMask(self):\n",
" inoutMask = (torch.randperm(28*28)+1).long()\n",
" hiddenMask = torch.randint(low=1, high=28*28, size=(self.hidSize,)).long()\n",
" MW = torch.zeros( (self.hidSize, 28*28) )\n",
" MV = torch.zeros( (28*28, self.hidSize) )\n",
" \n",
" MW[ hiddenMask.view(self.hidSize, 1) >= inoutMask.view(1, 28*28)] = 1.0\n",
" MV[ hiddenMask.view(1, self.hidSize) < inoutMask.view(28*28, 1) ] = 1.0\n",
" \n",
" return inoutMask, MW.to(device), MV.to(device)\n",
" \n",
" def nextMask(self):\n",
" self.currentMask = (self.currentMask + 1) % self.numMask\n",
" self.hidden.setMask(self.masks[self.currentMask][1])\n",
" self.output.setMask(self.masks[self.currentMask][2])\n",
" \n",
" def forward(self, x):\n",
" out = self.hidden(x)\n",
" out = self.output(out)\n",
" return self.loss(out, x)"
],
"execution_count": 0,
"outputs": []
},
{
"metadata": {
"id": "UVl0YWvH24Os",
"colab_type": "code",
"colab": {}
},
"cell_type": "code",
"source": [
"numMask = 1\n",
"net = MADE(numMask=numMask, hidSize=1000*numMask)"
],
"execution_count": 0,
"outputs": []
},
{
"metadata": {
"id": "L2UEIvqw3rXP",
"colab_type": "code",
"colab": {}
},
"cell_type": "code",
"source": [
"from torch.utils import data\n",
"net = net.to(device)\n",
"learning_rate = 1e-3\n",
"optimizer = torch.optim.Adam(net.parameters(), learning_rate)"
],
"execution_count": 0,
"outputs": []
},
{
"metadata": {
"id": "W7Cs8WMNm2pY",
"colab_type": "code",
"colab": {}
},
"cell_type": "code",
"source": [
"#@title\n",
"def log_sum_exp(value, dim=None, keepdim=False):\n",
" \"\"\"Numerically stable implementation of the operation\n",
"\n",
" value.exp().sum(dim, keepdim).log()\n",
" \"\"\"\n",
" # TODO: torch.max(value, dim=None) threw an error at time of writing\n",
" if dim is not None:\n",
" m, _ = torch.max(value, dim=dim, keepdim=True)\n",
" value0 = value - m\n",
" if keepdim is False:\n",
" m = m.squeeze(dim)\n",
" return m + torch.log(torch.sum(torch.exp(value0),\n",
" dim=dim, keepdim=keepdim))\n",
" else:\n",
" m = torch.max(value)\n",
" sum_exp = torh.sum(torch.exp(value - m))\n",
" if isinstance(sum_exp, Number):\n",
" return m + math.log(sum_exp)\n",
" else:\n",
" return m + torch.log(sum_exp)\n",
"\n",
"import math "
],
"execution_count": 0,
"outputs": []
},
{
"metadata": {
"id": "-zKY0_qO5G6e",
"colab_type": "code",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 1887
},
"outputId": "7c511a86-84db-4c8a-ed05-32c2f55ec184"
},
"cell_type": "code",
"source": [
"lo = 0.0\n",
"train_data = train_data.cpu()\n",
"valid_data = valid_data.to(device)\n",
"for it in range(100):\n",
" trainloader=torch.utils.data.DataLoader(train_data, batch_size=512, shuffle=True, num_workers=8)\n",
" for ii, data in enumerate(trainloader, 0):\n",
" ensamble = torch.empty(data.shape[0], numMask)\n",
" for mm in range(numMask):\n",
" oo = torch.sum(net(data.to(device)), dim=1)\n",
" ensamble[:, mm] = -oo \n",
" net.nextMask()\n",
" oo = log_sum_exp(ensamble, dim=1)\n",
" loss = -torch.mean(oo) + math.log(numMask)\n",
" optimizer.zero_grad()\n",
"\n",
" loss.backward()\n",
"\n",
" optimizer.step()\n",
" lo = 0.99* lo + (1.0 - 0.99) * loss.item()\n",
" print(\"loss: \", lo)\n",
" if it % 10 == 0:\n",
" mypr = torch.exp(torch.sum(torch.log(F.sigmoid(net.output(net.hidden(valid_data)))), dim=1))\n",
" ttt = torch.empty(valid_data.shape[0], numMask)\n",
" for itt in range(numMask):\n",
" ttt[:, itt]= -torch.sum(net(valid_data), dim=1)\n",
" net.nextMask()\n",
" \n",
" ll = -torch.mean(log_sum_exp(ttt, dim=1)) + math.log(numMask)\n",
" print(\"Test: \", ll.item())\n"
],
"execution_count": 0,
"outputs": [
{
"output_type": "stream",
"text": [
"loss: 143.94690040880047\n",
"Test: 173.7479705810547\n",
"loss: 148.80785201779014\n",
"loss: 137.6237066582583\n",
"loss: 126.85514776885935\n",
"loss: 118.91216422336373\n",
"loss: 113.3748374382782\n",
"loss: 109.5523535349712\n",
"loss: 106.82717074980077\n",
"loss: 104.86362155811926\n",
"loss: 103.34339353169632\n",
"loss: 102.17518981159895\n",
"Test: 102.1021957397461\n",
"loss: 101.19834867086368\n",
"loss: 100.40100644351512\n",
"loss: 99.72842793007072\n",
"loss: 99.12192684779541\n",
"loss: 98.59006481898864\n",
"loss: 98.12787023272493\n",
"loss: 97.69595993396311\n",
"loss: 97.3013363314196\n",
"loss: 96.92274058837734\n",
"loss: 96.58025262426617\n",
"Test: 98.01366424560547\n",
"loss: 96.24124026021715\n",
"loss: 95.91402458605164\n",
"loss: 95.60634374488606\n",
"loss: 95.28693758051995\n",
"loss: 95.0247063065077\n",
"loss: 94.72966699903313\n",
"loss: 94.43523291819007\n",
"loss: 94.198487038613\n",
"loss: 93.93182797789471\n",
"loss: 93.65815836531782\n",
"Test: 95.8995132446289\n",
"loss: 93.4562654205505\n",
"loss: 93.22593552522949\n",
"loss: 92.96747586153623\n",
"loss: 92.7546449706383\n",
"loss: 92.53700433905233\n",
"loss: 92.28302294409586\n",
"loss: 92.11236931963649\n",
"loss: 91.91971864734542\n",
"loss: 91.72862866007988\n",
"loss: 91.5165577441914\n",
"Test: 94.48977661132812\n",
"loss: 91.30143658563901\n",
"loss: 91.14631074891713\n",
"loss: 90.9941985604919\n",
"loss: 90.80584250710417\n",
"loss: 90.63118940532779\n",
"loss: 90.45821899486238\n",
"loss: 90.3060814370102\n",
"loss: 90.11896098386799\n",
"loss: 89.97564923918375\n",
"loss: 89.88048379492238\n",
"Test: 93.68921661376953\n",
"loss: 89.69570416293854\n",
"loss: 89.57605269726882\n",
"loss: 89.43928700030912\n",
"loss: 89.28387849201654\n",
"loss: 89.17998882268877\n",
"loss: 89.01942387805343\n",
"loss: 88.93084638834311\n",
"loss: 88.78789203808289\n",
"loss: 88.66322157851505\n",
"loss: 88.51464434698784\n",
"Test: 93.29536437988281\n",
"loss: 88.43985522142135\n",
"loss: 88.34141512413278\n",
"loss: 88.19704072985924\n",
"loss: 88.09672716379838\n",
"loss: 88.01566636490078\n",
"loss: 87.8778825460252\n",
"loss: 87.79738029429457\n",
"loss: 87.67471588673358\n",
"loss: 87.61200049005602\n",
"loss: 87.50351104212965\n",
"Test: 93.07516479492188\n",
"loss: 87.38635995017557\n",
"loss: 87.28241550316437\n",
"loss: 87.17824979023305\n",
"loss: 87.08991313625297\n",
"loss: 86.99682210643003\n",
"loss: 86.92442574127051\n",
"loss: 86.8565559728376\n",
"loss: 86.77801508605535\n",
"loss: 86.68285450020231\n",
"loss: 86.61577613446309\n",
"Test: 93.03553771972656\n",
"loss: 86.49192295606582\n",
"loss: 86.39768301912258\n",
"loss: 86.31396722213557\n"
],
"name": "stdout"
},
{
"output_type": "stream",
"text": [
"loss: 86.25418994299422\n",
"loss: 86.18110336366458\n",
"loss: 86.07295924843848\n",
"loss: 86.00837969202725\n",
"loss: 86.00086045866475\n",
"loss: 85.91371086144316\n",
"loss: 85.81255437119977\n",
"Test: 93.0876693725586\n",
"loss: 85.76141030313293\n",
"loss: 85.67209477475429\n",
"loss: 85.59822235947415\n",
"loss: 85.54632717983641\n",
"loss: 85.53318036622089\n",
"loss: 85.46411773320786\n",
"loss: 85.37625452722128\n",
"loss: 85.33080735494612\n",
"loss: 85.25627080837181\n"
],
"name": "stdout"
}
]
},
{
"metadata": {
"id": "7nBg172a8j6q",
"colab_type": "code",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 152
},
"outputId": "7616a317-ed6b-4c14-f087-0e99d87b418c"
},
"cell_type": "code",
"source": [
"#@title\n",
"rcParams['figure.figsize'] = 4,2\n",
"\n",
"plt.subplot(1, 2, 1)\n",
"_ = plt.imshow(F.sigmoid(net.output(net.hidden(train_data[141:142].cuda()))).view(28,28).cpu().data.numpy())\n",
"plt.subplot(1, 2, 2)\n",
"_ = plt.imshow(train_data[141:142].cuda().view(28,28).cpu().data.numpy())"
],
"execution_count": 0,
"outputs": [
{
"output_type": "display_data",
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAACHCAYAAADZeY9mAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
"text/plain": [
"<matplotlib.figure.Figure at 0x7effcdfcd550>"
]
},
"metadata": {
"tags": []
}
}
]
},
{
"metadata": {
"id": "N87D13bARKgd",
"colab_type": "code",
"colab": {
"base_uri": "https://localhost:8080/",
"height": 485
},
"outputId": "760173ef-8c09-4a75-bdb6-ed9585314a55"
},
"cell_type": "code",
"source": [
"def generateSample(net, n):\n",
" x = torch.zeros(n, 28*28).to(device)\n",
" dic = {}\n",
" for i in range(28*28):\n",
" loc = net.masks[net.currentMask][0][i] - 1\n",
" dic[loc.item()] = i\n",
" \n",
" for i in range(28*28):\n",
" idx = dic[i]\n",
" pr = F.sigmoid(net.output(net.hidden(x)))[:, idx]\n",
" x[:, idx] = torch.bernoulli(pr)\n",
" net.nextMask()\n",
" \n",
" return x\n",
"\n",
"net.nextMask()\n",
"digits = generateSample(net, 16)\n",
"\n",
"size = 16\n",
"rcParams['figure.figsize'] = 4*2,4*2\n",
"for i in range(1, size+1):\n",
" plt.subplot(4, 4, i)\n",
" _ = plt.imshow( digits[i-1].cpu().data.reshape(28,28) )"
],
"execution_count": 0,
"outputs": [
{
"output_type": "display_data",
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAAAdwAAAHUCAYAAACQ8JmYAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ngIILAIADFFwAAByg4AIA4AAFFwAAB/4/SNf56bfiplEAAAAASUVORK5CYII=\n",
"text/plain": [
"<matplotlib.figure.Figure at 0x7eff75acfd30>"
]
},
"metadata": {
"tags": []
}
}
]
}
]
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment